It is concluded that the ordinary hypothesis of rupture cannot be employed to predict safe range of alternating stress which can be applied to metal having a scratched surface; the theorem of minimum potential energy may be extended so as to be capable of predicting the breaking loads of elastic solids, if account is taken of increase of surface energy which occurs during formation of cracks; breaking load of a thin plate of glass having in it a sufficiently long straight crack normal to applied stress, is inversely proportional to square root of length of crack. It is shown that application of mathematical theory of homogeneous elastic solids to real substances may lead to error.